# DeepSeek-V2.5 (Sep 2024) vs Llama 3.2 3B

> DeepSeek-V2.5 (Sep 2024) is the stronger model overall, scoring 37.0 to 28.4 on the Noometry Index.

- Canonical page: https://noometry.com/compare/deepseek-v2-5-vs-llama-3-2-3b
- Last updated: 2026-10-10
- Shared benchmarks: 15

## Summary

- They share 15 benchmarks with published results for both. DeepSeek-V2.5 (Sep 2024) scores higher in 8 categories and Llama 3.2 3B in 0 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where DeepSeek-V2.5 (Sep 2024) leads 50.0 to 24.9.
- The biggest single-benchmark swing is BigCodeBench Instruct: 48.6% for DeepSeek-V2.5 (Sep 2024) and 23.4% for Llama 3.2 3B.

## Snapshot

| | DeepSeek-V2.5 (Sep 2024) | Llama 3.2 3B |
|---|---|---|
| Provider | DeepSeek | Meta |
| Noometry Index | 37.0 | 28.4 |
| Rank | 197 | 318 |
| Context | — | 131K |
| Input $/M | — | $0.05 |
| Output $/M | — | $0.33 |
| Weights | Open | Open |

## Coding

- DeepSeek-V2.5 (Sep 2024): 31.3 (#274)
- Llama 3.2 3B: 27.1 (#314)

| Benchmark | DeepSeek-V2.5 (Sep 2024) | Llama 3.2 3B |
|---|---|---|
| BigCodeBench Instruct | 48.6% | 23.4% |
| LMArena Coding | 1310 | 1098 |
| BigCodeBench Complete | 53.2% | 28.3% |
| Aider Polyglot | 17.8% | — |
| HumanEval+ | 83.5% | — |
| MBPP+ | 74.1% | — |

## Agentic & Tool Use

- DeepSeek-V2.5 (Sep 2024): —
- Llama 3.2 3B: 19.9 (#135)

| Benchmark | DeepSeek-V2.5 (Sep 2024) | Llama 3.2 3B |
|---|---|---|
| Berkeley Function Calling Leaderboard | — | 21.9% |
| BALROG | — | 10.1% |

## Reasoning

- DeepSeek-V2.5 (Sep 2024): 24.5 (#145)
- Llama 3.2 3B: 20.1 (#225)

| Benchmark | DeepSeek-V2.5 (Sep 2024) | Llama 3.2 3B |
|---|---|---|
| LMArena Hard Prompts | 1289 | 1095 |

## Math

- DeepSeek-V2.5 (Sep 2024): 33.3 (#177)
- Llama 3.2 3B: 29.9 (#213)

| Benchmark | DeepSeek-V2.5 (Sep 2024) | Llama 3.2 3B |
|---|---|---|
| LMArena Math | 1289 | 1126 |

## Knowledge

- DeepSeek-V2.5 (Sep 2024): 34.2 (#190)
- Llama 3.2 3B: 29.1 (#232)

| Benchmark | DeepSeek-V2.5 (Sep 2024) | Llama 3.2 3B |
|---|---|---|
| LMArena Expert | 1266 | 1090 |

## Multilingual

- DeepSeek-V2.5 (Sep 2024): 42.7 (#188)
- Llama 3.2 3B: 26.4 (#276)

| Benchmark | DeepSeek-V2.5 (Sep 2024) | Llama 3.2 3B |
|---|---|---|
| LMArena Non-English | 1273 | 1019 |
| LMArena Chinese | 1317 | 1017 |
| LMArena German | 1258 | 1056 |
| LMArena Russian | 1289 | 949 |
| LMArena French | 1288 | — |
| LMArena Japanese | 1228 | — |
| LMArena Korean | 1209 | — |
| LMArena Spanish | 1248 | — |

## Instruction Following

- DeepSeek-V2.5 (Sep 2024): 67.7 (#189)
- Llama 3.2 3B: 56.3 (#270)

| Benchmark | DeepSeek-V2.5 (Sep 2024) | Llama 3.2 3B |
|---|---|---|
| LMArena Instruction Following | 1280 | 1089 |

## Long Context

- DeepSeek-V2.5 (Sep 2024): 39.5 (#169)
- Llama 3.2 3B: 33.4 (#256)

| Benchmark | DeepSeek-V2.5 (Sep 2024) | Llama 3.2 3B |
|---|---|---|
| LMArena Longer Query | 1301 | 1101 |

## Writing & Preference

- DeepSeek-V2.5 (Sep 2024): 50.0 (#182)
- Llama 3.2 3B: 24.9 (#302)

| Benchmark | DeepSeek-V2.5 (Sep 2024) | Llama 3.2 3B |
|---|---|---|
| LMArena Text | 1294 | 1110 |
| LMArena Creative Writing | 1285 | 1094 |
| LMArena Multi-Turn | 1297 | 1105 |
| EQ-Bench Creative Writing | — | 595 |

## FAQ

### Is DeepSeek-V2.5 (Sep 2024) better than Llama 3.2 3B?

DeepSeek-V2.5 (Sep 2024) is the stronger model overall, scoring 37.0 to 28.4 on the Noometry Index.

### Is DeepSeek-V2.5 (Sep 2024) or Llama 3.2 3B better for coding?

DeepSeek-V2.5 (Sep 2024) scores higher on coding benchmarks: 31.3 versus 27.1 in the Noometry coding category.

### How many benchmarks do DeepSeek-V2.5 (Sep 2024) and Llama 3.2 3B share?

15 benchmarks have published results for both models. DeepSeek-V2.5 (Sep 2024) has 22 scored results on Noometry and Llama 3.2 3B has 18.